Re: [PATCH] gdb/python: add allow_pending parameter to gdb.Breakpoint constructor

[Simon Marchi <simark@simark.ca>]

On 10/22/26 9:54 AM, Andrew Burgess wrote:
> Tom Tromey <tom@tromey.com> writes:
> 
>>>>>>> "Tomáš" == Tomáš Hrnčiar <thrnciar@redhat.com> writes:
>>
>> Tomáš> Add a new allow_pending parameter to the gdb.Breakpoint constructor.
>> Tomáš> Setting the new parameter to False causes GDB to raise an error instead
>> Tomáš> of creating a pending breakpoint.  By default this parameter is True,
>> Tomáš> which allows pending breakpoints to be created, preserving the existing
>> Tomáš> behaviour.
>>
>> Hi.  Thanks for the patch.
>>
>> This is PR python/12816 so please add this tag to the commit message:
>>
>> Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=12816
>>
>> Tomáš> +  ** The gdb.Breakpoint constructor now takes an optional
>> Tomáš> +     'allow_pending' argument.  Setting it to False causes GDB to raise
>> Tomáš> +     an error instead of creating a pending breakpoint.
>>
>> I wonder about calling the new argument 'pending' instead.  I tend to
>> think that's a bit nicer since we already match the parameter and
>> attribute name for 'temporary'.
> 
> Hi Tom,
> 
> Full disclosure: I consulted with Tomáš about this patch before it was
> posted.
> 
> I suggested that the argument be called allow_pending rather than just
> pending as having:
> 
>   bp = gdb.Breakpoint("file.c:10", pending=True)
> 
> seemed (to me) that this was a request *for* a pending breakpoint, which
> clearly isn't what this actually does.  I felt that:
> 
>   bp = gdb.Breakpoint("file.c:10", allow_pending=True)
> 
> was much clearer about the intent.
> 
> Though, having just typed this, I realise that as 'True' is the default,
> you'll never actually write that, you'd instead only write:
> 
>   bp = gdb.Breakpoint("file.c:10", pending=False)
> 
> where reading this as a request *for* a non-pending breakpoint is
> actually fine.  So, I guess the case where the name 'pending' is (IMHO)
> confusing, is the case that you're less likely to write.
> 
> Still, I do prefer the fuller name, but if you'd rather it was changed,
> then that's also fine with me, I'm not that tied to it.

I like "allow_pending" for the reason you explained.

Simon